Protection circuits not protecting against overcharge?

[Fallingwater]

While attempting to fix my Photon Rex (which consistently overcharges its cell) I got the idea of adding a protection circuit from an ex-protected 18650 LiIon cell to the Rex.

So I fished it out of my "potentially useful junk" drawer, connected it...


(clickable)

...and the Rex cheerfully went on overcharging. I stopped it when the charging voltage reached 4.25V.

I assumed the protection circuit was bad, so I replaced it with one from a protected Trustfire 14500...


(clickable)

...and still the Rex overcharged. I stopped it at 4.23V.

I've given up hope with Ultrafire cells as they've disappointed me too many times; but the Trustfire 14500, I expected to be overcharge-protected. DX explicitly state it is in the page about it.

So what's the deal? Are the cells just completely unprotected from overcharge, or did I do something wrong? Aren't they supposed to open the circuit at exactly 4.2 volts?

 

[kramer5150]

I don't see how you can blame the manufacturer of the cell you tore apart. You are using their charge circuit with a Vin of 1.2V... and its from a battery so its going to vary greatly as the AA cell depletes. AFAIK Lithium Ion battery protection circuits are designed for charge voltages ~3.6V

A better way to monitor charge would be with a DMM to manually monitor voltage across the charging cells.

 

[jirik_cz]

The protection circuit is probably fine. Overcharge cut-off voltage is usually 4.25V+-0.05V or even 4.30+-0.05V.

 

[Fallingwater]

I don't see how you can blame the manufacturer of the cell you tore apart. You are using their charge circuit with a Vin of 1.2V... and its from a battery so its going to vary greatly as the AA cell depletes.

What does the Vin of the charging circuit have to do with anything? The important part is the Vout, that's what the circuit is supposed to protect from.

I've thrown together this testing rig:


(clickable)

The AMC driver and Cree LED are connected to the battery terminals on the protection circuit. They simulate a charging cell by drawing current (I used those because they were the first thing that crossed my mind; a suitable incandescent bulb, or just a resistor, would work fine).
By connecting my bench power supply to the contact terminals of the protection circuit I simulate a charger throwing energy in the cell.

Now, the protection circuit should trip if the voltage rises too much, right?
Well, the 18650 circuit, the 14500 one and even one from a cell phone battery cheerfully let through voltages well in excess of 4.5V.

So I reversed the flow, wiring them normally, with the power supply going to the battery terminals and the driver connected to the contact terminals.
No change. They all still blissfully let current pass at 4.5V; not only that: they didn't open the circuit even with the voltage going as low as 2.5V.

These protection circuits seem to be doing no protectin' at all to me. They may or may not protect against overload - I wasn't about to test for that - but they let the voltage go to levels that would be very seriously damaging to any LiIon cell.
If you can spot some fatal flaw in my testing I'm all ears; otherwise I'm starting to think most cells that we assume to be protected really aren't.

 

[kramer5150]

I was thinking maybe the protection circuit needs higher voltage than the 1.2V from the NiMH to run optimally.???

 

[Fallingwater]

You're probably not familiar with the Photon Rex. You connect it to a cell of up to 3V, and it steps up the voltage and uses it to charge its own LiIon cell. In its position inbetween the Rex and LiIon cell, the protection circuit never even knows the incoming energy's coming from the AA cell.

Aside from that, the circuits still didn't work in my test rig. I presume they do eventually cut out, but they still worked at 2.5V, and that's a dangerously low voltage for LiIon cells.
I'll conduct more testing to figure out when exactly they open, but if the other circuits are like the ones I'm testing I'm worried that they might all be very close to useless.

 

[shadowjk]

I wonder if it lets through 4.5V to battery because the current is still high, to maintain compatibility with pulsed chargers?

 

[qwertyydude]

I don't know if it affects it but the protection circuit from my Trustfires all cutout when 110% of the initial charge goes through. So if I have a depleted battery and charge at 1 amp, then only 1100 mah get charged. If I charge it at 500 then only 550 mah make it in. It's weird. My unprotected cells charge just fine but I usually have to charge a dead cell twice, at most 3 times at low currents.

 

[Fallingwater]

I don't know if it affects it but the protection circuit from my Trustfires all cutout when 110% of the initial charge goes through. So if I have a depleted battery and charge at 1 amp, then only 1100 mah get charged. If I charge it at 500 then only 550 mah make it in. It's weird. My unprotected cells charge just fine but I usually have to charge a dead cell twice, at most 3 times at low currents.

From what I've seen the circuits themselves don't have the brains to calculate how much energy has made it in.
I suspect it's your charger acting up rather than the protection. Mine certainly does this; with tired/old cells, or generally ones with relatively high internal resistance, it charges for exactly 66 minutes and 53 seconds (I'm sure about the minutes, not about the seconds; I'm going by memory here), thus charging the cell with 110% of the set charge rating. I have to restart the charge 2 or 3 times before it'll fill up the cell, though it'll go for as long as necessary without interruption once it reaches CV stage.
Do you by any chance own a Hyperion charger?

I wonder if it lets through 4.5V to battery because the current is still high, to maintain compatibility with pulsed chargers?

They seem to be letting it through no matter the current. I haven't yet seen what the circuits do if I put several amps through them; I hope they'll at least disconnect the cell then, otherwise they are completely useless and have no reason for existence.

 

[Gryloc]

Fallingwater,

You may have to check your test rig. First of all, if your LED/driver load is drawing lots of current current, then there will be a voltage drop in the test leads. Therefore, if you are taking your voltage reading from the display on the power supply (measured internally), you may see an incorrect voltage. The best thing to do is connect a DMM to the output of the protection circuit where the "battery" will go and not rely on the voltage reading from the power supply. With two 3ft 24AWG (maybe larger gauge) wires attached to my supply, I measured about 0.2V drop total at 1000mA of current.

There may be a voltage drop in the power leads, and there may be a voltage difference between the input and output of the protection circuit (due to resistance or due to the circuit consuming a little power). I believe that most protection circuits should cut off charging when it senses the voltage across the output hit 4.2V or so (someone mentioned the threshold be as high as 4.30V +/- 0.05V). The circuit does not care what the input voltage is, as long as the voltage it sees across the cell is in the acceptable range.

Your protection circuit may not like the strange load of the driver circuit and LED. There may be some weird high frequency switching happening in the driver circuit that throws off the protection circuit. Why not try testing a bank of 4 AA NiMH cells you have laying around. This way, these NiMH cells would better act like a li-ion cell, but 4.5V cannot do any harm if the protection circuit is faulty. I have had good luck with my protection circuits (the few I tested). I may have to test some more just to see...

So, try the DMM at the "battery". I am assuming you are relying on the voltage reading on the power supply. I cannot see your test rig in full by the pictures, and I didn't see a DMM attached anywhere on the output of the protection circuit.

-Tony

 

[mdocod]

PCBs are generally set to cutoff at ~4.30-4.35V in my experience. The reasons for this are many, but primarily, it is so that the cell will be fully charged when it is used in a charger that does not use a CV stage. Some are constant current all the way through the charge, and need to temporarily ramp the cell voltage up to ~4.30V in order to bring the cell to a full charge. In some cases, like that of Wolf-Eyes chargers and the OLD Pila chargers, it actually relied on the protection circuit to terminate the charge, and again, the charger did not have a CV stage, so needed to over-charge a bit, but at a fast current, to achieve a full charge. These charging methods are not ideal, but are found in many chargers and don't seem to be the source of any serious danger or issues provided they are used with the proper cell capacity ranges.

----

I noticed above that you tried to "simulate" a battery in the PCB with an LED instead of a battery, and you were still able to run up to 4.5V without it cutting off... This is interesting, and suggests to me one of the following things:
1. The PCBs are faulty (these are "fire" brands afteral)
2. The PCB requires an actual cell in the circuit, and the behavior of the cell (not an LED) to work correctly.
3. You may have damaged the PCBs upon removal from the cells.
4. The PCBs may not be what they claim to be. (incorrect spec listing on DX/KD is not unusual or anything)

You haven't really proven whether the protection works or not at this time, because you haven't tried "overcharging" an actual cell to ~4.35V. I don't really recommend this as it's not great for the cell...

Since you have a "retired" 18650, and a bench power supply, I would suggest using that for the testing... I would setup a test, ideally outside, where an exploding li-ion cell would not be a problem, and wire the retired 18650 to the PCB, then to the power supply, set the power supply to ~1.5A maximum, and ~4.4V maximum. If the PCB is working properly, the output current should suddenly drop to 0 and the voltage jump straight to 4.4V at the power supply side when the cell reaches ~4.35V, the cell should then settle within moments to under ~4.20V, with the cell no longer being charged.

Eric

PS: what I am suggesting could be dangerous, please use caution and be ready to deal with a fire if things go wrong, a bucket full of dirt and a respirator wouldn't hurt. If the charging reached ~4.37V and it still hasn't been terminated by the PCB, then you should manually terminate it and consider the PCB junk.

 

[Fallingwater]

You may have to check your test rig. First of all, if your LED/driver load is drawing lots of current current, then there will be a voltage drop in the test leads. Therefore, if you are taking your voltage reading from the display on the power supply (measured internally), you may see an incorrect voltage. The best thing to do is connect a DMM to the output of the protection circuit where the "battery" will go and not rely on the voltage reading from the power supply. With two 3ft 24AWG (maybe larger gauge) wires attached to my supply, I measured about 0.2V drop total at 1000mA of current.

I don't trust my power supply's built-in meters further than I could throw them. They are good for setting up the supply, but you really need precise meters for fine-tuning. Which is why I used three different ones in my measurements, and yes, I did check voltage at the cell.

The circuit does not care what the input voltage is, as long as the voltage it sees across the cell is in the acceptable range.

I tested this; the input voltage and the voltage across the cell are exactly the same. The circuits don't have any fancy regulation or anything going on, they (are supposed to) just pass current under normal conditions, and open the circuit under abnormal ones.

Your protection circuit may not like the strange load of the driver circuit and LED. There may be some weird high frequency switching happening in the driver circuit that throws off the protection circuit.

This is possible, I guess, though it sounds unlikely to me.

Why not try testing a bank of 4 AA NiMH cells you have laying around. This way, these NiMH cells would better act like a li-ion cell, but 4.5V cannot do any harm if the protection circuit is faulty.

This is a great idea. I'll rebuild the tester in this way as soon as I get home.
I'll use some big NiCD cells I have; this way I can run ungodly currents through the rig and test the circuit's overload protection as well, all without damaging any cells.

In some cases, like that of Wolf-Eyes chargers and the OLD Pila chargers, it actually relied on the protection circuit to terminate the charge

Man, talk about dumb design choices...

3. You may have damaged the PCBs upon removal from the cells.

All three of them?
I'm not exactly an electronic genius, but I know how not to destroy equipment I want to take apart. Removal is simple: no soldering is required, you just cut the strips holding the circuit to the cell and take it off.

You haven't really proven whether the protection works or not at this time, because you haven't tried "overcharging" an actual cell to ~4.35V.

This will change soon.

Since you have a "retired" 18650, and a bench power supply, I would suggest using that for the testing... I would setup a test, ideally outside, where an exploding li-ion cell would not be a problem, and wire the retired 18650 to the PCB, then to the power supply, set the power supply to ~1.5A maximum, and ~4.4V maximum. If the PCB is working properly, the output current should suddenly drop to 0 and the voltage jump straight to 4.4V at the power supply side when the cell reaches ~4.35V, the cell should then settle within moments to under ~4.20V, with the cell no longer being charged.

I'll do this if the NiMH/NiCD pack idea should turn out to be unfeasible for some reason.

be ready to deal with a fire if things go wrong, a bucket full of dirt and a respirator wouldn't hurt

I ain't overcharging on purpose LiIon cells in my home. If it comes to this I'll do it outside. Should the cell(s) blow I'll just take a few steps back and watch the fireworks.

 

[LuxLuthor]

Thanks for the PM.

Couple things I see as possibilities...and a few questions.

• 1) Not sure EXACTLY how you are checking the voltage. What quality of DMM, type of probes, exact placement location of leads when checking.

• 2) I don't start out believing anything from "El Cheapo" sites like DX or KD, since they are doing everything as cheaply as possible, cutting corners wherever possible, and no concern about quality or liability. Of course there can be exceptions, but that is my starting assumption.

Regarding this specific DX Trustfire link specs....I don't believe any of it....nor do I believe even if you got accurate specs that there would be any reliable QC from sample to sample. Reportedly Trustfire is better than Ultrafire...but that can change at any time, with no concern for QC. It wouldn't even surprise me to learn that if Trustfire actually has some concern for quality/reputation, there may be rejected batches/bins that are pawned off to places like DX.

Certainly the performance reputation of a cell has no bearing on the reliability/performance of a separately made PCB that are bought in lots. You have to stop and think how likely Trustfire thinks it will be for most consumers to actually check the accuracy of a PCB...vs...the cell's mAh & volt performance at various amp draw rates which is highly likely to be examined. That presumption will affect their motivation in doing any QC on the chips they accept, and/or paying a higher price for rejecting bad chips or requiring higher quality PCB's.
​

Here is a link to a single Li-Ion PCB from BatterySpace.com. Note near the bottom there is an image of PCB with specs that differ from the text on overcharge protection....and this is a reasonably reliable website!!! Is OC protection 4.275 +/- 0.045 V or 4.35 +/- 0.025 V ?? Is Over Current protection 4A or 9.5A ???
​

• 3) Depending on your DMM probe contact point, you may have some resistance lowering actual voltage at chip sensor.

• 4) You may have solder/heat induced malfuntion or deterioration of tracing &/or chip components in attaching your wires. (Admittedly unlikely).

• 5) Introducing the Photon ReX changes factors related to what a battery PCB will sense and react to. If you are going to test and verify how a Li-Ion PCB is working, you need to present actual current, voltage, resistances that are correlated with the PCB's design...including the battery resistance and CC/CV from a Li-Ion charger...or variable power supply.

 

[BoarHunter]

I second point 5 in the above post. What you supply the PCB with, is certainly not continuous current/voltage but pulsed and therefore may not be properly handled by the protection system.